Hi,

I have this specific "hum" and I can't figure out how to remove it. I have IzotopeRX, and the de-hum module doesn't do much. I can run spectral de-noise on it and get some of it removed. Obviously it affects the voice if I try to remove it completely. So, I'm hoping someone will be able to determine specifically, what it is and maybe tell me how to target just that noise.

https://www.dropbox.com/s/nbzyn4hteausfm1/Noise.wav?dl=0


Thank you
Kathy

If the file I downloaded is an accurate measure of the sample, it seems to be a very low level base noise that probably could be ignored, as -70 dB is pretty small. This kind of noise can be from normal electronic noise in the recording devices, background rumble in the venue of the recording, etc. My home, with even with a small and absorbent room, with everything electrical turned off, has a constant -52dB rumble much like this. Again, if the levels in the sample are the levels you are worried about, I don't think they are much of a problem. You are correct, de-hum is for a specific frequency and its harmonics; this is a broad-range random noise from near 0 to around 17khz, and that won't help. You could do a sample and remove using a regular noise removal tool if you feel it's necessary. The capture shows what looks to me like a typical random noise pattern that is, to some extent, always present in a recording.

The noise is broadband so a hum remover tool and would not work well in this scenario. 'Hum' removal tools typically work @ 50 or 60Hz and the pertinent harmonics.
The source file's noise is around -60dBFS, which should be inaudible for most playback situations, unless cranking the monitor volume to abnormal high levels or normalizing just the noise with no other content. The audio on the frequency spectrum graph was normalized to show the frequencies clearly.

So many tracks. What am I looking at here? I've got it in iZoRX10 Advanced and I can't figure out what I should be focussing on. Please explain.

https://www.dvinfo.net/forum/attachment.php?attachmentid=40029&stc=1&d=1671605502

OK, RX Spec Analyser giving -75, not bad at all. And now I "see" the other streams. Just what they are from your capture, like to know. But, it all depends on just what it is you're attempting to hunt down. Are you purely wanting to ascertain just where these low dbs are coming from? Curiosity? Is it impinging on your Project? Are you wanting something less than -75? Are you searching for what could be making this type of noise?

https://www.dvinfo.net/forum/attachment.php?attachmentid=40030&stc=1&d=1671606161

Nice project, but I need more of your premise for just what you are searching for - please?

We really need a sample of your problem audio in a file that also has the content you want to keep, not just the problem noise only. Too many unknowns with the file you have supplied.

Andrew

I used the AI-based process at audostudio.com for some problem tracks recently and it was pretty astonishingly good. You can process up to 20 minutes free as a sample. Very few settings but quite robust.

This was the test clip I did when first trying it out, where we weren't able to turn off the exceptionally noisy coolers behind a bar during a shoot. My post-production mixer couldn't match this with Izotope and has since used Audo on a number of projects, he says while it isn't always perfect, overall it is a huge time-saver.

https://www.dropbox.com/s/c22hdd03aq6zkt2/bar%20none%20audio%20test%202.mov?dl=0

We really need a sample of your problem audio in a file that also has the content you want to keep, not just the problem noise only. Too many unknowns with the file you have supplied.

Andrew

Indeed.

It would be like me being given a Ball of Sheep’s wool, a Knitting Pattern and needles and then being told to make a pair of socks. But, in this instance, I’ve got the finished product and am now being told to come up with the knitting pattern.

I would dearly love the original file and just what it is that needs to be done. Hence my questions above and the motivation from Kathy just what it is she’s wanting. To produce nicer sounding audio, or, to determine what is making that noise.

Oh and BTW, I can’t knit.

.

I have this specific "hum" and I can't figure out how to remove it.

Upload a full file with the voice. Very hard to ascertain anything from the download. Sure, a few of us would have a look at it for you.

Chris Young

What are the circumstances of the recording? That sounds more like mechanically generated noise than electronic noise? What are we actually listneing to? The sound of your preamp when nothing is plugged in? The sound of an empty room, via a mic .... or what? As everyone has said, it's extremely low, but then we've all normalised it to an annoying sound, but that's probably unrealistic? what are we listening to, recorded on what?

Kathy, I should have mentioned in my reply that if you are worried about the effect on your voice content from removing this very small signal, it should have no practical effect. It is a fairly even signal across the entire vocal range, so removing it would just, at most, reduce the amplitude of the content by a tiny amount. An ordinary noise-removal effect (sample and remove) would do the trick. Simple answer, try it and see. You are correct that noise removal can affect content --- if it is a significant part of the signal, and particularly if over a narrow frequency range; removing a spike like that leaves a "hole." But this is a different animal.

I have this specific "hum" and I can't figure out how to remove it. I have IzotopeRX, and the de-hum module doesn't do much.

Whatever the noise is, it's a combination of things.

Track #1 has two channels, both left and right tracks, which appear to be dual mono and appear to be identical to each other.

Track #2 is showing on both channels, but appears to be summed mono.

Tracks 3 & 4 are purely White Noise. Track #3 showing on both left and right channels, and track #4 only on the left channel.

If I had the job of processing this "audio" I would totally lose tracks #3 and #4 as they are purely "noise". I would ditch Track #2 as it appears to be a summed mono version of what is on Track #1.

Finally, looking at Track #1's two channels, I could clean this up very quickly with Sound Soap, I think? Not having the associated audio with this 'noise' makes it hard to come to a definitive conclusion, but I have cleaned up this sort of noise quite successfully in the past.

Chris Young

BIAS SoundSoap Pro 2 Promo/Tutorial Video - YouTube

Hi All,

Thanks for all the replies. I think I'm obsessing over nothing. I think this is the room noise that I'm hearing (the hum of the fluorescent bulbs, perhaps). I know you asked me for details but in the interest of everyone's time, I don't think it's worth discussing it, as I can't hear it when I don't listen on my monitoring headphones.
It actually turns out that I won't be using any of this recording. I'm not happy with the content.

Thank you everyone who jumped it to help.

Kathy

Kathy - lots of people spent time on this for you, and you don't think it's worth discussing? That's a little rough, if you don't mind me saying so.

The thing to remember is that you haven't fixed it, diagnosed it, or done anything to prevent it happening on a track that IS important.

We tried.

For what it's worth - the audio format you recorded it in is very unusual? All thos summed and empty tracks? Maybe you are recording on something a bit strange? If the noise worried you enough to join a forum and post - the worry will still be there?

Happy Christmas

C'est la vie, as they say.

Anyway... have a happy holiday season everyone.

Chris Young

Kathy - lots of people spent time on this for you, and you don't think it's worth discussing? ……Happy Christmas

This post has a low SN Ratio.

This post and the low SN ratio is what I've experienced with previous Kathy posts. Marginal audio (understandably as it's a legit cause for posting about a problem), not enough of it to really work with, and it feels like everything is secret squirrel. You wouldn't even know what it was about / what the project was in the end. At least she has a copy of Izotope RX these days, to give credit where due.

Kathy, we're always happy to help if you're open to it. Just be real with us and it's all good. We're all very friendly over here (as you will have noticed).

Andrew

..as I can't hear it when I don't listen on my monitoring headphones.

FWIW: Listening with headphones (especially tightly coupler over the ear) will generally make any background noise/hiss/hum much more noticeable.
and
Background noise tends to intrude more when gain is increased to compensate for mic placement that is increasingly distant from the talent/desired source.

I used the AI-based process at audostudio.com for some problem tracks recently and it was pretty astonishingly good. You can process up to 20 minutes free as a sample. Very few settings but quite robust.

Thanks for the tip about Audostudio. I tried varied recordings and I found that it's amazingly good for speech. Apparently the AI recognizes what is speech, and gates it ON. That's as opposed to recognizing unwanted noise and gating it OFF.

For example, a recording w/ speech + bg noise is processed very well. A recording with singing and noise is not processed correctly, apparently because singing does not have the same sound pattern as speech, so a lot of the singing is partially removed. A recording w/ speech, desired SFX, and continuous background noise (e.g. from the storage medium) is not processed correctly ... the speech is retained OK, but the SFX is mostly removed, along with the background noise.

Bottom line: for working with dialog stems this is probably great. But for restoring mixed recordings (what I'm working on lately) this is the wrong tool. Certainly overkill for Kathy's current super-secret project.

I'm sorry. I'm recording on Sound Devices mixpre-3. I set it up so I can record each channel separately. That's why have all of these channels when I open the file. I just delete all of them leaving just the one that has the voice in the Center channel.
I don't have any issues with this set up. When the room noise is low the recording is pretty clean to my ear.

Here is a snippet of the recording with the voice:
https://www.dropbox.com/s/dx5lzzmb8otaf5a/Clip.wav?dl=0

Thank you, and again apologies for dropping this topic without any further input from me and not appreciating the time an effort everyone put into trying to help and solve this for ME.

Kathy

Thank you, and again apologies for dropping this topic without any further input from me and not appreciating the time an effort everyone put into trying to help and solve this for ME.

Kathy

No worries from me, glad to see you back here and to have that snippet to analyze. I now understand the multi-track scenarios - it did throw me a bit. Now if my approach is incorrect and happy to be corrected.

My initial thoughts are that the low dB for the voice will kind of "mask" any noise floor that's there. I'd say that your MAX recorded VO is 50% of what you are presenting here:
https://www.dvinfo.net/forum/attachment.php?attachmentid=40033&stc=1&d=1672207190

Once I GAINED your clip to -10dB I could clearly "hear" the noise floor. I don't think you've got enough to hear that rumbling noise with your cans. This is what I can achieve by adding +6dB GAIN:

https://www.dvinfo.net/forum/attachment.php?attachmentid=40034&stc=1&d=1672207932

And now with adding a further +6dB GAIN - ie +12dB over the original:

https://www.dvinfo.net/forum/attachment.php?attachmentid=40035&stc=1&d=1672207932

And now clearly reveals rumbling at -45dB:

https://www.dvinfo.net/forum/attachment.php?attachmentid=40036&stc=1&d=1672207932

So, to recap, your sample does have some room noise rumbling, I've lived with less-than-perfect Audio - oh yes! But, if my attempts to analyze your sample has any useful response on your initial question, that you're looking for noise, then I would have to suggest it sounds like room noise. You're not hearing over your cans as a result of this masked by the initial low recording dB.

HTH!

OK, I've now used Noise reduction on that rumbling. I've now got a cleaner Signal and a Noise floor of -85dB! I like this:

https://www.dvinfo.net/forum/attachment.php?attachmentid=40037&stc=1&d=1672209565

Here's my DropBox link:

https://www.dropbox.com/s/qf1yzgp80jlb63a/Kathy%20Full%20treatment%201a.mp4?dl=0

I have this specific "hum" and I can't figure out how to remove it. I have IzotopeRX, and the de-hum module doesn't do much. I can run spectral de-noise on it and get some of it removed. Obviously it affects the voice if I try to remove it completely.

OK…. I’ve now listened and re listened comparing yours and mine. I’m more than convinced that the noise is coming from PC fans and being amplified by your room and surroundings. Reducing this by 50%, from -45dB to -85dB allows your Signal to win the day, and smother the noise floor. If I’m wrong about your PC fans then you’d have to look deeper and maybe consider removing any electrical forces impinging on your recording.

As Zappa said: "Sometimes yah just gotta walk away from a project!" - I'm done . . .

Further treatment was to remove clipping > Post-treatment Analogue Audio Warmth > Light touch of reverb.

Here's that treatment:

https://www.dropbox.com/s/dkqgvbcwsl90sel/Kathy%20Nopise%20treatment%20Post%20Plugs1a.wav?dl=0

I'm done!

O...K...

One last thought, in any way, is your connection to your pc being used via the System Audio? Or are you set up using the drivers for that Pre3? What are the drivers for it?

Graham, what do mean by "cans'? I'm not connected to PC at all. What drivers are you talking about?

“Cans” is a slang term for headphones.

Your second part of your feedback is related to the Pre3. You record to it and then you upload it to your PC? Yes?

I’ve noted you’ve made no comment on my analysis nor my treatment and my methodology.

“Cans” is a slang term for headphones.

I have not heard that much since the 70's.

“Cans” is a slang term for headphones.

I have not heard that much since the 70's.

Old habits . . .

“Cans” is a slang term for headphones.

Your second part of your feedback is related to the Pre3. You record to it and then you upload it to your PC? Yes?

I’ve noted you’ve made no comment on my analysis nor my treatment and my methodology.

Graham, I'm in transit so I can't listen to anything

“Cans” is a slang term for headphones.

I have not heard that much since the 70's.

I'm pretty sure I heard it somewhere (most likely a regional radio station) in the 90s. Can anyone better that?

Andrew

Try this shop: https://theradioshop.co.uk/collections/headsets-cans

Thanks for reminding me of my age and my aging Sennies! Ah, nostalgia, it’s not what it used to be.

Nostalgia was so much better in the old days!

Andrew

https://playbutton.co/why-are-headphones-called-cans/

When a kid we played with the tin-can-and-string-telephones.

Cans is still very common in the UK - more now for 'headsets with mics'. In theatre it's ettiquette to say "off cans" when you take them off so people know you're gone and won't respond any more.

In the studios as far back as I can remember, we always used the one syllable word “Cans” quicker and cooler than ‘Headphones.’

Cheers.

Hi,

I have this specific "hum" and I can't figure out how to remove it. I have IzotopeRX, and the de-hum module doesn't do much. I can run spectral de-noise on it and get some of it removed. Obviously it affects the voice if I try to remove it completely. So, I'm hoping someone will be able to determine specifically, what it is and maybe tell me how to target just that noise.

A quick clean-up attempt. With a bit more time, could be improved. Download from here:

Chris Young

https://www.sendspace.com/pro/dl/w3k5s5

A quick clean-up attempt. With a bit more time, could be improved.

Chris Young

Chris, nice job. I took the liberty of running your sample through RX Waveform stats. What do you think?:


https://www.dvinfo.net/forum/attachment.php?attachmentid=40043&stc=1&d=1672467470

Yes Graham. I like that yours has more bottom end and sounds fuller. I found when I first tried giving it a bit more bottom end, it muffled the recording somewhat. Your recording made me revisit my first attempt.

To get to the result of version #1 that I uploaded, I just duplicated track #1 with the file on it to another track, track #2. I ran noise reduction on track #1. I then ran noise reduction on track #2.

Then I inverted the phase on track #2. This gave me 100% signal cancellation with track #1. No sound audible as the two tracks cancelled one another out. I then offset track #2 by three (3) samples to the right. This brought my tracks back to being audible. Obviously, though, sounding thin due to the opposing phase cancellation of track #2 vs track #1. On the master bus, I then added a DeEsser at 5506Hz with a -30dB threshold.

On this #02 version upload, again on the master bus, I added a +7.5dB boost @ 167Hz plus boosted the overall gain to peak at about -3dB. In other words, to levels I'm more than likely to use during an edit. All of it done 100% in Vegas Pro 17.

The inverted phase cancellation with offset method is something I've used for years when having to try to extract reasonable audio out of noisy ambient room mics for example.

Chris Young

Version #2 can be found here:
https://www.sendspace.com/pro/dl/6ejypm

Yes Graham. I like that yours has more bottom end and sounds fuller. I found when I first tried giving it a bit more bottom end, it muffled the recording somewhat. Your recording made me revisit my first attempt.
Well, thank you for saying. I always add gain to start with. A bit like in video seeing how much Gain I can add without video noise. So, I add GAIN in stages until I got a well-represented fuller SN ratio that I could use, then off to the Races with iZoRX10 Adv.

To get to the result of version #1 that I uploaded, I just duplicated track #1 with the file on it to another track, track #2. I ran noise reduction on track #1. I then ran noise reduction on track #2.

Yes, I cut my Audio teeth within the VP and SoundForge packages. Using much the same approach as yours. Dupe Tracks; Inversions and the NR package. However . . . . iZotopeRX came along and I haven't done any Audio correction within VP or SF for the last 5 to 10 years, only iZo - It is pure VooDoo!

Again I captured the Waveform statistics in iZo and produced this graphic for comparison:

https://www.dvinfo.net/forum/attachment.php?attachmentid=40048&stc=1&d=1672486153

Understood Graham.

I also run iZotope RX10, Celemony Melodyne, Capstan (audio de-wow, fantastic for old tape sources). Along with a host of other audio tools and apps. Most history doco producers have to run a comprehensive set of audio and video tools these days.

Resolve is my main NLE these days. One of the reasons I'm still using Vegas is that I'm still working on a doco series that was started in Vegas ten years back because it was then the only totally video agnostic NLE in terms of mixing any and every type of video format back then. This series is still slowly moving forward. It's for the government. It's the official history series of the Australian RAN. The history of the navy since 1788.

This project is using footage of every conceivablevideo frame rate and dimensions from film and tape and archived film on a variety of files from MPEG1/2, AVI, MXF, MOV. PAL, NTSC, SD and HD etc, etc. Along with all manner of old audio archive material off film, tape and ancient audio files. The audio from most of these archives is incredibly poor in a lot of cases. Some of the audio dates back to WWI. The round tripping of audio in and out of other programs, then finally into Pro Tools for the final mixdown was becoming a pretty onerous task.

To simplify cleaning up all this audio in one spot while actually editing the footage, I decided to arm Vegas to the teeth with whatever audio tools were required and that could work with it. Over the years I have built a very comprehensive set of VST plugs that work within Vegas, mainly iZo and Waves. Having to clean up everything from ancient newsreels and ancient political radio speeches has been a great challenge. I'm very old school. Started working with audio with the BBC back in the '60s. Nothing digital, all analogue, no readouts, no repeatability, no saving presets. All we had were VU and phase meters. You tweaked until it sounded "right". Enjoyable though.

For potent and very quick and configurable dialogue noise reduction, it's very hard for me to go past Waves Clarity Vx Pro for my work these days. For me, this has been a game changer in terms of speed. Not to mention simplicity. Dial it as you listen.

Chris Young

Waves Clarity Vx Pro: The New Standard in Noise Reduction for Voice - YouTube

Chris - KUDOS!! Your experience far exceeds both my humble requirements and needs. In the short time I've been involved with Video, 20 years now, I've learned to try, absorb and share, and this during the tremendous technology upheaval in our Industry. For me, it's been exciting and financially rewarding, adding to which, coming across individuals, like yourself, who are willing and able to communicate their feedback.

In Kathy's last sample, the first thing that came to my mind was that it was suffering from a lack of microphone proximity. The second thing was that the content from the talent was a load of faffed about bollocks. I'm not at all surprised she found the audio to be unusable.

There are some things digital tools just can't fix.

Andrew

In Kathy's last sample, the first thing that came to my mind was that it was suffering from a lack of microphone proximity. Indeed. Always get the mic as close as possible to source. But, again, we start from what we have now.
The second thing was that the content from the talent was a load of faffed about bollocks. Nicely put. However, as pros we can’t necessarily get too dismissive of what was said, although it was a little ……There are some things digital tools just can't fix.

Andrew Correct.

In Kathy's last sample, the first thing that came to my mind was that it was suffering from a lack of microphone proximity. The second thing was that the content from the talent was a load of faffed about bollocks. I'm not at all surprised she found the audio to be unusable.

There are some things digital tools just can't fix.

Andrew
Not to disparage your's and Graham's efforts, but it seems what you did made the lack of proximity more apparent. There's also a subtle distortion that I'm hearing in both of the clean up samples. I think the distortion is inherent in the recording. Just more noticeable with the gain added. I gave a quick try just using gain then spectral denoise and the proximity issue wasn't as noticeable. I'll post a sample.

https://drive.google.com/file/d/157-f5wQXgXpo5jeApnVnXvmZmn8Ml81C/view?usp=share_link

Yes, good work, nice repair, nice presence. No noticeable artefacts. That's the winning and telling point.

Chris Young